Today the analysis of customer Panama WhatsApp Number List in natural language (ticketing, satisfaction surveys, feedback, your comments under this article for example) is often done … by hand, by humans who devote a lot of time to it. ‘even more considerable when the opinions to be analyzed are numerous: with its artificial intelligence algorithms, Phedone makes it possible to extract from several thousand opinions an intelligible, exploitable and agnostic value in a few minutes. In addition, there is often a dimension missing from analyzes that send customer reviews in large numbers and in natural language: the identification of synergies between customer reviews. Phedone responds to these issues with a triptych:
Micro analysis of each verbatim to extract its substance (if we did it by hand, we would speak here of “qualification” of opinions) The macro analysis of all the verbatim documents which makes it possible to identify trends, weak signals, the main themes addressed (by hand we would speak here of “synthesis”) Synergistic analysis: through the prism of collective intelligence, Phedone crosses opinions to identify group trends (by location, product line, profile, etc. of each customer), the perception games specific to entities, Trades, etc. We have a first product in production based on many homemade technological bricks – the subject of a patent application – or open source.
What issues does Phedone address
The voice of the patient (hospitals, research laboratories, etc.) where in their medical journeys, patients are rarely encouraged to express themselves through open questions – and even when they are, this data is not or little exploited: in this segment there is therefore also a real stake in the data collection phase to allow the improvement of the patient journey. At this stage, Phedone is mainly used by customers in Telecommunications and Retail: these sectors are particularly interesting for us because the volumes of customer reviews are very important – their feedback helps us a lot to test the relevance of the analyzes of the solution and its resilience to the volumes injected.
The voice of the customer where we help in particular with ticketing services for IS users of large companies. First success with these users: we succeeded in reducing the time necessary for the qualification of a ticket from 15-20min to 100ms. Our solution was also tested as part of a citizen consultation conducted by the Grand Est Region: in addition to the time savings made for the analysis of citizen opinions, Phedone intervened here as an agnostic tool for the qualification of opinions. policies. In a few minutes we were able to draw up a panorama of subjects of interest to citizens for the Region.
Where is Phedone today and what next
The genesis of Phedone was made around a question of the world of research: how to scale the results of research on collective intelligence using digital tools? To continue to improve our solution, we have opened Phedone to everyone, do not hesitate to try it, your feedback is of interest to us! In addition, the engines of our solution are unsupervised and Phedone adapts to data reception: our second objective is to deploy engines and models specific to Businesses while developing the interoperability of Phedone with SI flagship of our customers. To this end, we now devote 30% of our investments to R&D.
[Jonathan F.]: I trained as an engineer which ended with end-of-study projects on the application of AI for the segmentation of brain tumors. In 2018, I moved to the corporate side, to SAP then to other software publishers where, no longer to do with brain tumors, I was doing product management. I then set up an association that brings together engineers and researchers in artificial intelligence: DeepNet. Our goal was to give the general public access to educational content on AI with a “positive impact” scope, that is to say by orienting our work on societal and environmental subjects. Subsequently, I set up a first company – JustAI – training and providing AI services to help start-ups